For a hand tool, varying the dimensional value of the selection parameters leads to different model variants. All the variants are not able to provide postural comfort and safety during work. This study suggests a method to identify ergonomically suitable hand tool variants that provide postural safety and comfort. Ten hand hoe variants used in the study are determined by varying the dimensional values of handle length, handle diameter and blade weight. The dynamically varying postures during soil loosening activity using each variant are identified from the video of hand hoe use. A sequence of image processing operations is used to extract, identify and classify postures from the video. The video-based data of ten women agricultural workers with ten hand hoe variants are classified and analysed for posture risk based on OWAS. The hand hoe variants are ranked based on postural safety and comfort. Thus, this paper presents an approach for the selection of hand tool variants considering posture risk.